Depressive symptoms, including negative self-focused emotions (e.g., sadness, guilt), make trauma- exposed Veterans vulnerable to developing depressive pathology and substance use disorders (SUD). Both conditions are in turn linked to persistent executive and functional impairments. A common marker of both depression severity and addictive behavior is anhedonia, i.e., reduced ability to seek and experience pleasure/ healthy rewards, with evidence of reduced neural activity in the reward centers of the brain. Such reward sensitivity alterations are thus likely to play a critical role in promoting depressive and addictive pathology in trauma-exposed Veterans. A precise understanding of the neurocognitive mechanisms supporting reward- based decision-making, and how mood is integrated into these processes, is therefore needed to: a) understand the contribution of these risk factors to current and future psychopathology, b) help detect and predict treatment needs/outcomes for Veterans with such complex clinical profiles. Bayesian models can offer a powerful quantitative account of these intricate mechanisms by disentangling: a) learning processes (prediction of reward likelihood in the environment), and b) action selection strategies (choosing an action based on those predictions). This computational framework can help better delineate how low mood impacts both the prediction of rewards, and the integration of these predictions into decision strategies. Moreover, given that existing treatments for trauma-exposed Veterans focus on reducing anxiety symptoms, there is a significant need for developing behavioral treatments that can more thoroughly target depressive symptoms associated with low reward responsiveness, such as anhedonia and substance use. Computationally based assessment of reward processing alterations may be particularly useful and timely for providing new treatment targets and directions for improving such treatment outcomes for trauma-exposed Veterans. To address these questions, we propose to use computational modeling and neuroimaging to identify precise affective neurocognitive predictors of psychopathology and behavioral treatment response in recently deployed young (age 20-40) trauma-exposed Veterans. This project will use Bayesian modeling applied to the analysis of reward-based decisions, with baseline dependent measures of brain circuit activity derived from functional magnetic resonance imaging (fMRI), and experimental manipulation of mood, to delineate precisely a) how sad mood may bias the learning or strategic adjustments guiding reward-based decisions in Veterans, b) the degree to which these affect-driven computational biases relate to depression and problem substance use symptoms in this population, and c) the degree to which these computational markers can predict Veterans' response to a cognitive behavioral treatment targeting depressive symptoms associated with trauma, including guilt and anhedonia. In other words, can computational methods be useful in identifying low reward sensitivity in the depressed brain, and in directing early on Veterans with such markers towards more appropriate and effective treatments? The outcomes of this study will identify more precise neurocognitive predictors of reward responsiveness, affective clinical profile, and treatment response, and provide a cognitive rationale for developing more targeted and effective interventions for Veterans. This CDA-2 award will facilitate the applicant's training in the utilization of computational and neuroimaging techniques 1) to understand the role of affect-driven cognitive biases in promoting depressive and substance use symptoms, and 2) to predict treatment response and long-term clinical course. Such training will in turn facilitate the applicant's long-term career goals to become a productive and successful VA clinical scientist, whose work focuses on improving early detection and effective treatment of psychopathology in Veterans.